(this was originally under
Quadword)
I think this probably isn't adequate. I mean, I know what "unit," "computer memory or storage," "four," and "word" all mean, but I still don't know what "quadword" means (at least, I don't think I do). Isn't that interesting? --LS
That is because no one has explained how many bytes are in a word. In IBM/370 Assembler a word has 32 bits or 4 bytes. A doubleword is the length of 2 words, thus 64 bits or 8 bytes. I was entering something about this under word with a one word instruction as an example and found someone disagreeing with me---I give up. Rose Parks
problem is that different machines have different word sizes
OK, different problem: if you use "word" without further ado,
many people reading the article won't know that you mean something different from what is ordinarily meant by "word."
I found this link on a page with byte...what was I to think?
RoseParks
The article asserts that now word usually means 16 bits but I think it still means "most efficient unit size on some particular architecture" or "architecture defined sequential packing of bits". What do others think? --
drj
I think the author is showing his/her IBM PC x86 bias by saying a word is 16 bits.
I have been around enough architectures that I consciously don't decide how big a word is, until someone tells me.
Somewhen in the past, drj was completely right, word meant the machine's "native" bit width, i.e. registers would always contain a word. But I think the trend to call 16 bits a word always, even outside of the now quite rare 16-bit beasts, is definitely there. This may be due to the wintel juggernaut, but it may also be due to there being a need to call 16 bits (and 32 bits, etc.)
something. Word is certainly not the best choice, but most people that are not chip designers seem to find it adequate.